The present invention enables the on-line measurement of semiconductor wafers or integrated circuits with high lateral resolution. In particular, the invention can be used to estimate the thickness profile of a sample consisting of a multilayer thin film on a substrate. The invention can also be used to estimate thermal properties of a sample, or quantities that influence these properties such as the ion implantation dose in semiconductor wafers. Moreover, the present invention can determine properties of a sample non-destructively and without contact with the sample, and can be used in on-line conditions on product wafers.
There are also many other possible industrial applications of the present invention. Whenever thin films or layers are deposited on a sample, or whenever properties of a sample need to be measured on short length scales, this invention can provide a wide range of information about the physical properties of the sample, such as mechanical or thermal properties crucial to the performance of the final product.
The non-destructive measurement of physical properties on micron, sub-micron, or nanometer length scales with optical techniques has been proposed in a wide variety of contexts.
For example a method for measuring the thickness or sound velocity of micron or sub-micron thin films has been proposed by J. Tauc et al. (see U.S. Pat. No. 4,710,030). This relies on the excitation of short wavelength stress pulses in a sample with ultrashort duration pump optical pulses and the detection of stress-induced changes in the optical constants of the sample using delayed probe optical pulses, in particular by measuring the changes in intensity of the probe beam reflected from the sample.
However, this method fails when the optical constants of the sample do not vary significantly with stress or strain in the sample at the wavelength of light used.
A similar method, that can be used to get round this limitation, was proposed by O. B. Wright and K. Kawashima (see Phys. Rev. Lett. vol. 69, 1668-1671 (1992)). It is based on the angular deflection of a probe beam arising from surface motion. Stress or strain pulses are always accompanied by motion of the sample surface or sample interfaces, and so this method can also be used to measure the thickness or sound velocity of thin films. This method, simultaneously sensitive to both phase and intensity changes in the light reflected from the sample, requires that the ultrashort optical pump and probe beams should be focused to slightly different positions at the sample surface. However, this method has the disadvantage of being very sensitive to the alignment of the pump and probe beams, in particular to the pump and probe optical spot separation on the sample. Moreover, the resolution is limited by the pointing stability of the laser used and by mechanical vibrations.
Another related method to generate and detect stress pulses with ultrashort optical pulses, proposed by T. F. Crimmins et al. (see Appl. Phys. Lett. vol. 74, 1344-1346, 1999) and H. J. Maris et al. (U.S. Pat. No. 5,864,393), relies on a grating technique making use of two pump beams at oblique incidence. This method, simultaneously sensitive to both phase and intensity changes in the light reflected from the sample, requires that two pump beams are simultaneously focused to the same spot on the sample at non-normal incidence.
However, measurements at normal incidence are highly desirable because the spots are circular and also because a single microscope objective can then be used to focus all the optical beams onto the sample, facilitating alignment and the obtention of a small optical spot size. Moreover, this grating technique requires several optical fringes to be produced on the sample at the same time, and this constitutes another reason why the method is not suitable for the highest lateral resolution measurements with the smallest optical spot sizes.
Another example of the use of optical techniques for the measurement of physical properties on short length scales is the use of thermal waves in the measurement of film thickness or ion implantation dose. A. Rosencwaig et al. (see Appl. Phys. Lett., vol. 46, 1013-1015, 1985), W. L. Smith et al. (see Nucl. Instrum. and Methods Phys. Res. Sect B, vol. B21, 537-541, 1987) and J. Opsal et al. (see U.S. Pat. No. 5,074,669) propose using chopped CW light beams to periodically excite thermal waves in a sample and to probe the resulting change in optical constants with a probe beam. This method can be used to probe to depths into the sample to within the thermal diffusion length, and is typically used for probing up to chopping frequencies in the 10 MHz region. However, at higher frequencies above 1 GHz, the method becomes impractical because the periodic temperature changes, dependent on the energy deposited in the sample by the excitation beam in one chopping cycle, decrease rapidly with increasing frequency. Higher frequency measurements are useful, because the thermal diffusion length, proportional to 1/{square root over ( )}f (f the chopping frequency), becomes smaller and the signals then become more sensitive to thermal properties of the sample in the near-surface region. When films of nanometer order in thickness are deposited on a substrate, it is, however, desirable to use such higher frequencies or equivalent short time scale pulsed optical measurements. Such short time scale pulsed optical measurements were proposed by C. A. Paddock and G. Eesley (see Opt. Lett., vol. 11, 273-275, 1986) using ultrashort pump and probe optical pulses. By monitoring the changes in optical intensity of a probe pulse reflected from the sample, it was possible to obtain a measure of the temperature change of the sample in the near-surface region in response to a pump pulse, and from the characteristic decay time of this temperature change estimate the thermal diffusivity of the sample. However, this method relies on the coupling of the optical reflectivity of the sample to temperature changes. If the optical reflectivity of the probe beam does not depend significantly on the temperature changes of the sample, the method fails.
Many optical interferometers have been proposed to measure the ultrafast changes in the properties of a sample in response to excitation by optical pulses on picosecond time scales. Such measurements have the advantage of being simultaneously sensitive to both phase and intensity changes in the light reflected from the sample, allowing a maximum of information to be extracted. If the optical reflectivity does not depend significantly on temperature or stress, there remains the possibility of monitoring temperature changes or the presence of stress pulses from changes in the optical phase. In addition, interferometric techniques are very sensitive, and interferometers can be easily calibrated to give quantitative information about, for example, the amplitude of the motion of the sample surface.
One example of an interferometric technique is time-division interferometry, in which an ultrashort optical probe pulse and an ultrashort optical reference pulse separately interact with a sample. For example, if the reference pulse is set to always arrive before a corresponding pump pulse and the probe pulse is set to always arrive after a corresponding pump pulse, the effect of the pump pulse on the sample can be determined by interfering the probe and reference pulses at a detector. Such a configuration, based on a Mach-Zehnder interferometer, was proposed by L. Sarger et al. (see J. Opt. Soc. Am. B, vol. 11, 995-999, 1994). This interferometer was designed for measurements in transmission through a sample. The Mach-Zehnder configuration is not a common path configuration, because it requires the probe and reference beams to travel completely different optical paths. It is thus is susceptible to unwanted noise sources such as mechanical vibrations or temperature changes.
Another configuration based on a Sagnac interferometer was proposed by M. C. Gabriel et al. (see Opt. Lett., vol. 16, 1334-1335, 1991) for measurements in transmission through a sample. This time-division interferometer is common path and is thus very stable with respect to mechanical vibrations or temperature changes. The same configuration could be used for measurements in reflection from a sample by placing the sample in the place on one of the mirrors in the Sagnac ring. However, it is not suitable for measurements at normal incidence. Moreover, the theory for the response of samples to ultrashort optical pulse excitation is simplified for the case of probing at normal incidence, rendering the data analysis much easier.
Another configuration based on a Sagnac interferometer was proposed by M. J. LaGasse et al. (see Appl. Phys. Lett., vol. 54, 2068-2070, 1989) for measurements in transmission through a sample. However this configuration is also not suitable for measurements at normal incidence. Moreover, this configuration is not suitable for opaque samples.
Another configuration based on a Michelson interferometer was proposed by J. Fieldler and J. W. Wagner (see Rev. Prog. Quant. Nondestr. Eval., vol. 16, 1579-1584, 1997) and C. J. K. Richardson et al. (see J. Opt. Soc. Am. B, vol. 16, 1007-1015, 1999) to monitor stress pulses in a sample generated with ultrashort optical pump pulses. This configuration is relatively stable but is, however, not suitable for measurements at normal incidence. Moreover, the probe and reference beams are not incident on the same point on the sample, and so errors can arise when lateral variations in optical reflectivity of the sample are present.
Another configuration based on a Michelson interferometer was proposed by O. B. Wright (Japan Patent Application Laid-Open (kokai) No. 5-172739, 9th July, 1993) to monitor stress pulses in a sample generated with ultrashort pump optical pulses through, in particular, the effect of stress-induced surface vibrations on the optical phase. This Michelson configuration does allow measurement at normal incidence. However the design is not common path. It is thus susceptible to unwanted noise sources such as mechanical vibrations or temperature changes.
Sagnac interferometers have also been designed for use with CW light beams. An example of this is disclosed in the U.S. Pat. No. 5,894,531 (J. J. Alcoz). The basic elements of the interferometer, making use of two arms and two beam splitters, is suitable for measurements at normal incidence and for detecting ultrasonic waves. However, working with CW light beams is not suitable for measurements on short time scales, in particular on picosecond time scales. This approach is therefore not suitable for the investigation of the physical properties of the sample on short length scales in the direction perpendicular to the sample surface.
Methods proposed by H. J. Maris (U.S. Pat. Nos. 5,706,094, 5,864,393) and by H. J. Maris and R. J. Stoner (U.S. Pat. Nos. 5,748,317, 5,748,318, 5,844,684 and 5,959,735) involve refinements of the original method by J. Tauc et al. (U.S. Pat. No. 4,710,030) for the monitoring of physical properties of materials. These methods involve the use of ultrashort optical pulses for the measurement of mechanical and thermal properties through, for example, variations in optical reflectivity and optical phase.
However, the methods described in these patents that are sensitive to optical phase variations; i.e., the beam deflection technique and the grating technique, are not suitable for the highest lateral spatial resolution measurements involving both pump and probe pulses focused to diffraction limited spots. The beam deflection technique, as previously mentioned, requires an offset between the pump and probe pulses and hence results in a degradation in lateral resolution.
In addition, the stability of the beam deflection technique is limited by the pointing stability of the laser used and by mechanical vibrations. The other method, the grating technique, requires several optical fringes to be produced on the sample at the same time using two pump beams at oblique incidence. For the reasons already explained above this technique is not suitable for measurements with the highest lateral spatial resolution.
One more proposed method in these patents and by O. B. Wright (Proceedings of the Ultrasonics Symposium, 1995, pp. 567-575) involves obtaining a better resolution than the diffraction limit using optical near-field techniques, such as the use of a tapered optical fiber. However, this method, because of the small aperture involved, implies the use of very low levels of optical power, and it therefore has the disadvantage of a much degraded signal-to-noise ratio.
As described above, conventional apparatuses for measuring physical properties of samples have various problems.
In the present invention, the measurement of properties of materials on short length scales, from 0.1 nm to 100 xcexcm, can be achieved by the use of short optical pulses, typically from 0.01 ps to 1 ns in duration, to monitor changes in physical properties of a sample induced by pump optical pulses, such as the generation and propagation of short wavelength stress pulses or thermal waves.
The effective measurement of such changes in physical properties of a sample with high stability using short optical pulses requires a simple and robust apparatus. In order to be able to apply the technique to a wide range of samples it is advantageous to use interferometry, inherently simultaneously sensitive to both phase and intensity changes of light reflected from the sample, allowing a maximum of information to be extracted. In order to obtain high spatial resolution, stability and to simplify the data analysis, it is advantageous to use a single lens at normal incidence to monitor the ultrafast changes in physical properties of a sample induced by pump optical pulses. This configuration is also desirable from the point of view of improved lateral resolution because the pump and probe light can be focused to a diffraction limited circular spot. At the same time it is highly desirable to use a common path design in order to minimize the effects of spurious mechanical vibrations or temperature changes.
The present invention provides an apparatus for measuring physical properties of a sample based on the use of ultrashort optical pulses that aims to satisfy all these requirements simultaneously.
An object of the present invention is to provide an apparatus for measuring the physical properties of a sample by optically monitoring the response of the sample to illumination by ultrashort optical pulses.
Another object of the invention is to provide such an apparatus that does not make mechanical contact with the sample and that can be non-destructive.
Still another object of the invention is to provide such an apparatus that can measure physical properties of the sample on short length scales in the direction perpendicular to the sample surface.
Still another object of the invention is to provide an apparatus that can measure physical properties of a sample with a horizontal resolution of the order of 1 micron or less. The lower limit of the resolution is determined only by the optical diffraction limit.
Still another object of the invention is to provide such an apparatus that can measure physical properties of a sample by optically monitoring the response of the sample to illumination by ultrashort optical pulses using a design that allows measurements that are simultaneously sensitive to both phase and intensity changes of the light reflected from the sample and, at the same time, allows measurements that are limited in lateral resolution only by the optical diffraction limit with a circular spot.
Still another object of the invention is to provide such an apparatus that can measure physical properties of a sample by optically monitoring the response of the sample to illumination by ultrashort optical pulses using a design that is insensitive to the effects of spurious mechanical vibrations or temperature changes.
Still another object of the invention is to provide such an apparatus that can measure physical properties of a sample by optically monitoring the response of the sample to illumination by ultrashort optical pulses with the optical measurement beam incident on the sample at normal incidence, in order to facilitate data analysis, alignment and the obtention of a high lateral resolution.
Still another object of the invention is to provide such an apparatus that can measure physical properties of a sample by optically monitoring the response of the sample to illumination by ultrashort optical pulses, in which diffraction limited light is focused in a circular region on the sample, in order to obtain a high lateral resolution.
Still another object of the invention is to provide such an apparatus in which it is possible to arrange in addition for both the pump and measurement beams to be incident on the sample at normal incidence using the same means for focusing for both.
Still another object of the invention is to provide such an apparatus that can spatially map the sample in the lateral direction and can measure physical properties of the sample corresponding to in the in-plane direction in the sample.
Still another object of the invention is to provide such an apparatus that can measure stress pulses generated by an optical pump pulse in samples without the requirement of a variation of the optical constants with stress.
Still another object of the invention is to provide such an apparatus that can measure temperature changes in samples without the requirement of a variation of the optical constants of the sample with temperature.
Still another object of the invention is to provide such an apparatus that can measure physical properties of very thin films.
Still another object of the invention is to provide such an apparatus that can measure samples consisting of one or more layers or consisting of a material with a buried permanent inhomogeneity or inhomogeneities.
Still another object of the invention is to provide such an apparatus that can measure samples made up of opaque, transparent or semitransparent parts or a combination of these.
In order to achieve the above objects, the present invention provides an apparatus which can measure the maximum amount of information about the optical response of a sample on short length scales when illuminated by ultrashort optical pulses, while at the same time satisfying the requirements for insensitivity to the effects of spurious mechanical vibrations or temperature changes and the requirement for high lateral spatial resolution down to the optical diffraction limit, ease of alignment, and ease of data analysis.
This invention features an optical generation and detection system for measuring the physical properties of a sample. There is a coherent or partially coherent radiation source for providing a pulsed measurement beam usually consisting of a periodic train of ultrashort optical pulses, having a duration of 0.002 ps to 2 ns. A typical example of such a radiation source is a mode-locked solid-state laser of high repetition rate.
There is also a radiation source for providing a pump beam usually consisting of a periodic train of ultrashort optical pulses, usually derived from the same radiation source as the measurement beam, having a duration of 0.002 ps to 2 ns.
The interferometer section of the apparatus is based on a Sagnac design. The interferometer section of the apparatus features two beam splitters used for the measurement beam. At the first beam splitter the measurement beam is split into two beams of different states of polarization, the probe beam and the reference beam. There is in general at this stage light mixed in with these beams that is not used in the measurement. The probe beam travels in the arm with the longer optical path between the first beam splitter and the second beam splitter, and the reference beam travels in the arm with the shorter optical path between these two beam splitters. The probe beam and the reference beam are recombined at the second beam splitter. The word xe2x80x98armxe2x80x99 here refers to a path in the interferometer that is traversed by the reference and probe pulses at distinct different times. The difference in the optical path lengths of the two arms is chosen to be longer than the spatial extent, in the direction of propagation of the light, of the optical pulses in the reference and probe beams. Equivalently, the difference in length of the arms xcex94L is chosen to be longer than cxcex94t, where c is the speed of light in the medium of the arms and xcex94t is the duration of the optical pulses making up the probe or reference beams. If the two arms correspond to different media a similar condition applies.
This recombined beam is directed onto the sample at normal incidence, passing, for example, through a quarter wave plate and a focusing lens. This recombined beam is reflected from the sample. The probe and reference beams then traverse opposite paths after passing a second time through the second beam splitter in such a way that the probe beam travels in the arm with the shorter optical path and the reference beam travels in the arm with the longer optical path of the interferometer. There is in general at this stage light still mixed in with these beams that is not used in the measurement. The probe beam and the reference beam are then recombined at the first beam splitter.
This beam then traverses the first beam splitter so as to produce a first output beam containing the probe and reference beams separate from the measurement beam. A means for introducing a known phase difference between corresponding optical pulses in the reference beam and the probe beam at a photodetector is used, typically a quarter wave plate placed in the path of the first output beam or before the first beam splitter. The known phase difference is chosen to be not equal to 0xc2x1180I degrees, where I is an integer; in order to ensure that the measurements are simultaneously sensitive to changes in both the phase and the intensity of the light reflected from the sample. The phase difference between the probe and reference beams is typically chosen to be xc2x190 degrees. This first output beam then passes through a linear polarizer. The probe and reference beams, or part of them, then interfere at a photodetector or photodetectors. The signal or signals from these photodetectors are representative of the changes induced in the sample by the pump beam. Because the probe and reference beams travel along identical optical paths, apart from the small time interval representing the optical transit time through the system, the interferometer is common path and insensitive to spurious mechanical vibrations or temperature changes.
A pump beam is directed onto the sample from a chosen direction, that can be collinear with the probe and reference beams when incident on the sample. The intervals in arrival times between the pump, probe and reference pulses can be chosen or varied at will, to produce measurements as a function of these time intervals.
The illuminated region of the sample can be scanned over the sample, or the region illuminated by the probe and reference beams can be scanned with respect to the region illuminated by the pump beam, in order to spatially map the signal or signals simultaneously dependent on ultrafast changes in both phase and intensity of the light reflected from the sample.
The optical pulses comprising the pump beam excite, in general, carriers, temperature changes and stress pulses in the sample. These stress pulses can be, for example, longitudinal waves, shear waves, surface waves or interface waves, including, for example, Lamb waves, Rayleigh waves, Love waves or Stoneley waves. These carriers, temperature changes or stress pulses give rise to mechanical motion of the surface of the sample or of interfaces in the sample and also to changes in optical constants of the sample. The purpose of the present invention is to provide an improved system for measurements that are simultaneously sensitive to both phase and intensity changes of the light reflected from the sample, and hence to monitor signals related to the changes induced by the pump beam. These signals can be analyzed to determine physical properties of the sample such as thickness, sound velocity, elastic constants or thermal properties, or quantities that influence these properties such as ion implantation dose.
It is also possible to monitor ejectant from the sample or irreversible transformation of the sample in the case when the optical pulses interact with the sample destructively, and to use this, for example, to measure the plasma dynamics or other parameter of the irreversible transformation.